Bitcoin Research in Princeton CS

Continuing our post series on ongoing research in computer security and privacy here at Princeton, today I’d like to survey some of our research on Bitcoin. Bitcoin is hot right now because of the recent run-up in its value. At the same time, Bitcoin is a fascinating example of how technology, economics, and social interactions fit together to create something of value.

Our Bitcoin work started with a paper by Josh Kroll, Ian Davey and me, about the dynamics and stability of the Bitcoin mining mechanism. There was a folk theorem that the Bitcoin system was stable, in the sense that if everyone acted according to their incentives, the inevitable result would be that everyone followed the rules of Bitcoin as written. We showed that this is not the case, that there are infinitely many outcomes that are stable yet differ from the written rules of Bitcoin. So the rule-following behavior that we currently see is at best stable in the weaker sense that if everyone else is following the rules (and no one mining entity has too much power) then deviating from the rules will cost you money.

Beyond this, we have built a better understanding of the “political economy” of Bitcoin—how the Bitcoin community governs itself to keep the system operating well, despite the lack of a central authority and despite the complicated issues around the theoretical stability of the protocol. The ultimate goal of this line of work is to understand how Bitcoin is likely to deal with challenges in the future, and whether there are feasible changes that could improve the governance of Bitcoin.

Since then, we have started several more Bitcoin-related projects. My faculty colleague Arvind Narayanan (who joined us last year) as well as several more students are working on Bitcoin, and the pace has accelerated. We’re building tools to track and diagnose the behavior of the peer-to-peer network that Bitcoin participants use to spread information about what is happening. We’re looking at the dynamics of mining pools, in which a group of miners cooperate to spread the risk inherent in the mining process. We’re considering new types of double-spending attacks and how participants can defend against them.

Let me highlight one current project: we’re designing a decentralized prediction market using the Bitcoin protocol. Prediction markets enable participants to trade “shares” on potentially any event with well-defined outcomes, such as a presidential election or sporting events. The market prices of these shares can be interpreted as the probability of the event occurring. Prediction markets offer societal benefits because of this ability to accurately aggregate the wisdom of crowds. Decentralization can improve prediction markets in various ways including robustness to closure (see Intrade), greater expressivity in defining markets and outcomes, and potentially lower fees leading to more accuracy in pricing unlikely events.

There are two main difficulties: first, how can a pair of anonymous participants trade shares without a trusted party to facilitate the transaction? Second, who will arbitrate the outcome of events? This is far trickier than it sounds—even for outcomes that are completely uncontroversial, some entity or group of entities must be entrusted with the authority to declare the outcome, and there must be checks to prevent them from abusing their power. It turns out that the contract-signing capability and the consensus mechanism of Bitcoin or a Bitcoin-like system enable us to find solutions to these problems, and that is the crux of our research. This is a collaboration between Princeton researchers and soon-to-be-CITP-fellow Joseph Bonneau, Jeremy Clark at Concordia, and Andrew Miller at UMD.

The analogy is often made that Bitcoin will do to money what the Internet did to communications. If that is the case, many, many interesting and useful designs that use Bitcoin as an underlying protocol are waiting to be discovered. It’s an exciting time to be doing research in this area.